Household energy storage emergency power supply priority intelligent scheduling method and system fusing user behavior small sample learning

Abstract

The invention relates to the technical field of household energy storage, and relates to a household energy storage emergency power supply priority intelligent scheduling method and system fusing user behavior small sample learning. The method comprises the following steps: firstly, collecting small sample data of a user, constructing a model based on meta-learning, and obtaining a personalized user behavior recognition model through meta-training and fine tuning; based on the model, outputting a user dependence degree score of each electric device, combining with a preset device emergency demand degree score, calculating an initial score by using a two-dimensional priority evaluation function, and sorting the initial scores; during emergency power supply, information such as energy storage residual electric quantity is acquired in real time, a dynamic adjustment model is constructed according to the initial score, and a dynamic priority sequence and a power distribution scheduling instruction are generated; executing the instruction, collecting user correction operation data, and updating the model through small sample incremental learning. According to the invention, the emergency power supply personalized adaptation degree and the energy utilization efficiency can be improved, and the core power demand interruption risk of the user is reduced.

Description

Technical Field

[0001] This application relates to the field of residential energy storage technology, and specifically to a method and system for prioritizing residential energy storage emergency power supply by incorporating small-sample learning of user behavior. Background Technology

[0002] Currently, residential energy storage systems, as an important component of distributed energy, can guarantee the core electricity needs of households in emergency scenarios such as power outages. The rationality of their emergency power supply dispatch directly affects user experience and electricity safety. Existing residential energy storage emergency power supply dispatch technologies have the following core pain points that have not yet been effectively resolved: Rigid priority scheduling and lack of personalized adaptation: Existing technologies mostly adopt preset fixed priority rules (such as default lighting > sockets > kitchen appliances), without considering users' personalized power consumption habits. For example, home office users have higher priority for computer emergency needs than lighting, while elderly users have the highest priority for medical equipment. Fixed rules cannot adapt to the core needs of different users.

[0003] Relying on a large amount of historical data, poor adaptability to new users / low data scenarios: Existing scheduling solutions based on user behavior analysis mostly use deep learning models, which require collecting several months of user electricity consumption history data for training. However, in scenarios such as new occupants and renters, the amount of user electricity consumption data is small (usually only 1-2 weeks), which makes it impossible to train the model effectively and achieve personalized scheduling.

[0004] In emergency situations, there is a lack of dynamic adjustment mechanisms: existing technologies mostly schedule based on initial priorities, without dynamically adjusting priorities in conjunction with real-time information such as changes in the remaining energy storage capacity and users' temporary power needs. This may result in problems such as failure to ensure power supply to core equipment before the energy storage capacity is exhausted, or insufficient power supply to high-priority equipment.

[0005] High model update costs: Most existing models are trained in batch mode. When users’ electricity usage habits change or the scene changes, a large amount of data needs to be collected again to train the model, which has a large computational cost and cannot achieve real-time adaptation.

[0006] Few-shot learning techniques (such as meta-learning) can achieve rapid model adaptation with limited data, providing a technical path to address the aforementioned pain points. However, currently, no technology combines few-shot learning with priority scheduling of emergency power supply for residential energy storage, particularly lacking a complete solution for key issues such as feature extraction from user behavior data, rapid adaptation of personalized models, and dynamic priority adjustment in emergency situations. Therefore, there is an urgent need for an intelligent priority scheduling method for emergency power supply of residential energy storage that integrates user behavior few-shot learning to achieve personalized and dynamic emergency power supply scheduling in scenarios with limited data, thereby improving the emergency service capabilities of residential energy storage.

[0007] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0008] To provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended as a general commentary, nor is it intended to identify key / important components or describe the scope of protection of these embodiments, but rather as a prelude to the detailed description that follows.

[0009] This disclosure provides a method and system for intelligent scheduling of emergency power supply priority for residential energy storage that integrates small-sample learning of user behavior. It solves the problem of user behavior modeling under small-sample data, improves the personalized adaptability and energy utilization efficiency of emergency power supply, and reduces the risk of interruption of users' core power needs in emergency situations.

[0010] In some embodiments, the method includes: Collect small sample data of users, including device power consumption time series data, user operation behavior data and scene label data, perform data cleaning, time series alignment and feature extraction, and construct a small sample feature set of user behavior; A user behavior few-sample learning model based on meta-learning is constructed. An initial meta-model is trained through a meta-training phase, and the initial meta-model is fine-tuned using the few-sample feature set in a fine-tuning phase to obtain a personalized user behavior recognition model. Based on the personalized user behavior recognition model, the user dependence score of each electrical device is output. Combined with the preset emergency demand score of the device, the initial priority score of each device is calculated through a two-dimensional priority evaluation function and then sorted. In emergency power supply mode, the remaining energy storage capacity, output power limit and power demand of each device are obtained in real time. Based on the initial priority score, a dynamic adjustment model is constructed to generate dynamic priority sequence and corresponding power allocation scheduling instructions. The scheduling instructions are executed, and user correction operation data is collected during the power supply process. The personalized user behavior recognition model is updated through small-sample incremental learning.

[0011] Preferably, the feature extraction includes temporal feature extraction and semantic feature extraction. The temporal features include peak power consumption periods, start-stop intervals, and power fluctuation coefficients. The semantic features are obtained through one-hot encoding of scene tags and sequence encoding of user operation behaviors.

[0012] Preferably, the meta-learning adopts a model-independent meta-learning algorithm, and the meta-training stage uses a multi-household common electricity consumption behavior dataset for training.

[0013] Preferably, the two-dimensional priority evaluation function is:

[0014] in, Rate the priority. Rate user dependency. Score the equipment's emergency demand level. This is the weighting coefficient, with a value range of 0.4–0.6.

[0015] Preferably, the constraints of the dynamically adjusted model include: The total output power of energy storage shall not exceed the rated power; The power supply of each device shall not be less than its minimum operating power; The remaining energy storage capacity should not be lower than the emergency protection threshold; When the remaining energy storage capacity is lower than a preset threshold, a priority upgrade mechanism is activated, supplying power only to high-priority devices that are within the previous preset proportion.

[0016] Preferably, the few-sample incremental learning only updates the local parameters of the personalized user behavior recognition model, without retraining the entire model.

[0017] In some embodiments, the intelligent scheduling system for priority emergency power supply of residential energy storage, which integrates user behavior small-sample learning, includes: The small sample data acquisition unit is used to collect device power consumption time-series data, user operation behavior data, and scene label data. The edge intelligent computing unit is used to run the user behavior few-sample learning model to realize personalized user behavior recognition and dynamic priority calculation. The energy storage dispatch execution unit is used to receive dispatch instructions and execute power allocation and power supply switching; The feedback interaction unit is used to collect user correction operation data and feed it back to the edge intelligent computing unit. The power switching unit is used to quickly switch to energy storage power supply mode when the power grid fails.

[0018] Preferably, the edge intelligent computing unit is equipped with a lightweight meta-learning inference framework, which supports the deployment and inference of model-independent meta-learning algorithms; The energy storage scheduling execution unit includes an energy storage converter control module and a battery management system interaction module.

[0019] In some embodiments, the intelligent scheduling device for priority emergency power supply of residential energy storage, which integrates user behavior few-sample learning, includes a processor and a memory storing program instructions. The processor is configured to execute the intelligent scheduling method for priority emergency power supply of residential energy storage, which integrates user behavior few-sample learning, when running the program instructions.

[0020] In some embodiments, the computer-readable storage medium stores a computer program that, when executed by a processor, implements the intelligent scheduling method for priority emergency power supply of residential energy storage based on small-sample learning of user behavior.

[0021] The present disclosure provides a method and system for intelligent priority scheduling of emergency power supply for residential energy storage that integrates user behavior small-sample learning, which can achieve the following technical effects: Strong adaptability to scenarios with limited data: This invention only requires 1-2 weeks of small data samples to complete personalized model adaptation, solving the modeling problem of lacking a large amount of historical data in scenarios such as new users and renters, and effectively improving scenario adaptability.

[0022] High precision of personalized scheduling: It integrates the dual-dimensional assessment of user dependence and equipment emergency demand, and the scheduling strategy is tailored to the user's actual core needs, effectively improving the power supply guarantee level of core equipment in emergency situations and enhancing user satisfaction.

[0023] Improved emergency power supply efficiency: The dynamic adjustment mechanism combines the remaining energy storage capacity to optimize priorities in real time, avoids wasting energy storage capacity, ensures that core equipment is powered until the energy storage SOC reaches the protection threshold, and extends the duration of emergency power supply.

[0024] Efficient and low-cost model updates: The model updates local parameters by using incremental learning with small samples, without retraining the entire model, which greatly reduces update time and computational overhead, and enables real-time adaptation to changes in user behavior.

[0025] Flexible deployment and easy to promote: The edge intelligent computing unit adopts a lightweight design and can be integrated into existing residential energy storage systems without large-scale hardware modifications. It is compatible with different brands and models of residential energy storage devices, reducing promotion costs.

[0026] The above general description and the description below are exemplary and illustrative only and are not intended to limit this application. Attached Figure Description

[0027] One or more embodiments are illustrated by way of example with reference to the accompanying drawings. These illustrations and drawings do not constitute a limitation on the embodiments. Elements having the same reference numerals in the drawings are shown as similar elements. The drawings are not to be scaled. And wherein: Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram illustrating the implementation methods provided in this disclosure. Figure 3 This is a schematic diagram of the system structure provided in the embodiments of this disclosure; Figure 4This is a schematic diagram of the device structure according to an embodiment of the present disclosure. Detailed Implementation

[0028] To provide a more detailed understanding of the features and technical content of the embodiments of this disclosure, the implementation of the embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for illustrative purposes only and are not intended to limit the embodiments of this disclosure. In the following technical description, for ease of explanation, several details are used to provide a full understanding of the disclosed embodiments. However, one or more embodiments may still be implemented without these details. In other cases, well-known structures and devices may be simplified in their depiction to simplify the drawings.

[0029] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this disclosure described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.

[0030] Unless otherwise stated, the term "multiple" means two or more.

[0031] In this embodiment of the disclosure, the character " / " indicates that the objects before and after it are in an "or" relationship. For example, A / B means: A or B.

[0032] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0033] The term "correspondence" can refer to an association or binding relationship. The correspondence between A and B means that there is an association or binding relationship between A and B.

[0034] Example 1 like Figure 1As shown, a user behavior-based small-sample learning-based intelligent scheduling method for emergency power supply priority in residential energy storage is proposed. Addressing the technical pain points of existing residential energy storage emergency scheduling methods, such as reliance on fixed priority rules, the need for large amounts of user behavior data to train models, and poor adaptability to new user / household type changes, this invention proposes a user behavior small-sample learning model based on a meta-learning framework. This model can accurately identify personalized electricity consumption habits using only 1-2 weeks of user electricity consumption data, dynamically generate emergency power supply priorities, and iterate and adjust priorities in real time by combining remaining energy storage capacity and equipment emergency demand. This invention solves the problem of user behavior modeling under small-sample data, improves the personalized adaptability and energy utilization efficiency of emergency power supply, and reduces the risk of interruption of users' core electricity needs in emergency situations. It is suitable for small-data residential energy storage application scenarios such as new residents, renters, and renovation of old communities, and has significant practical value and promising prospects for widespread application.

[0035] Specifically, the method includes: S1: Collect small sample data of users, including device power consumption time series data, user operation behavior data and scene label data, perform data cleaning, time series alignment and feature extraction, and construct a small sample feature set of user behavior; S2: Construct a user behavior few-sample learning model based on meta-learning. Train the initial meta-model through the meta-training stage, and fine-tune the initial meta-model using the few-sample feature set in the fine-tuning stage to obtain a personalized user behavior recognition model. S3: Based on the personalized user behavior recognition model, output the user dependence score of each electrical device, combine it with the preset device emergency demand score, calculate the initial priority score of each device through the two-dimensional priority evaluation function, and sort them. S4: In emergency power supply status, the remaining energy storage capacity, output power limit and power demand of each device are obtained in real time. Based on the initial priority score, a dynamic adjustment model is constructed to generate a dynamic priority sequence and corresponding power allocation scheduling instructions. S5: Execute the scheduling instruction and collect user correction operation data during the power supply process, and update the personalized user behavior recognition model through small sample incremental learning.

[0036] As a refinement of the above embodiment, step S1 involves collecting small sample data of users within 1-2 weeks, including device power consumption time-series data (start-stop time, running time, and real-time power of each device), user operation behavior data (manual device start-stop records, power adjustment records, and device priority start-up records in emergency situations), and scene tag data (weekdays / holidays, home hours / away hours, and special scene tags (working from home, infant care, elderly care)). The small sample data is then cleaned, time-series aligned, and features extracted (including time-series feature extraction and semantic feature extraction; time-series feature extraction includes peak device power consumption periods, start-stop intervals, and power fluctuation coefficients; semantic features are obtained through one-hot encoding of scene tags and sequence encoding of user operation behaviors), constructing a small sample feature set of user behavior.

[0037] As a refinement of the above embodiment, step S2 constructs a meta-training-fine-tuning two-level meta-learning framework and adopts the Model-Agnostic Meta-Learning (MAML) algorithm. In the meta-training stage, an initial meta-model is trained using a multi-household general electricity consumption behavior dataset (containing electricity consumption behavior data of different house types (one-bedroom / two-bedroom / three-bedroom) and different user groups (single youth / three-person family / elderly family)). In the fine-tuning stage, a small sample feature set of user behavior collected within 1-2 weeks is input into the initial meta-model, and the model parameters are adapted through a small number of gradient descent iterations to obtain a personalized user behavior recognition model.

[0038] As a refinement of the above embodiment, step S3 constructs a trained user behavior recognition model to output a user dependence score for each electrical device (representing the degree of user dependence on the device in an emergency, with a score range of 0-1). Combined with the device emergency demand score (which is determined based on device functional attributes, with medical devices, lighting devices, and communication devices having higher base scores than entertainment devices and non-essential kitchen devices), a two-dimensional priority evaluation function is constructed. The two-dimensional priority evaluation function is as follows: ,in Rate the priority. Rate user dependency. Score the equipment's emergency demand level. This is a weighting coefficient, with a value ranging from 0.4 to 0.6. It can be dynamically adjusted based on user scenario tags to calculate and sort the initial priority scores of each device.

[0039] As a refinement of the above embodiment, step S4 collects the remaining power (SOC), output power limit and real-time power demand of each device of the household energy storage system in real time. Based on the initial priority score, a dynamic adjustment model is constructed with the goal of maximizing the satisfaction of core needs and efficient utilization of energy storage power, and a dynamic priority sequence and corresponding power allocation scheduling instructions are generated.

[0040] The constraints of the dynamic adjustment model include: the total output power of energy storage does not exceed the rated power, the power supply of each device is not lower than its minimum operating power, and the energy storage SOC is not lower than the emergency protection threshold (10%). When the energy storage SOC is lower than 20%, the priority upgrade mechanism is activated, and only the top 30% of high-priority devices are supplied with power.

[0041] As a refinement of the above embodiment, step S5 monitors the user's corrective operation behavior and equipment operating status in real time during the emergency power supply process according to the generated scheduling instructions, feeds the corrective data back to the personalized user behavior recognition model, completes the incremental update of the model parameters, and realizes the continuous optimization of the priority scheduling strategy.

[0042] It should be noted that: This invention employs Meta-Learning (MAML) through a two-tier architecture of meta-model training and small-sample fine-tuning. It achieves personalized user behavior recognition model adaptation using short-term user electricity consumption data, breaking through the bottleneck of large data dependence in existing technologies.

[0043] An evaluation function is constructed by integrating user dependence and device emergency needs. Dynamic weight coefficients are used to adapt to different user scenarios, avoiding the problem of rigid fixed priorities.

[0044] A multi-constraint adjustment model is constructed by combining the remaining power of energy storage and the real-time power demand of equipment. When power is insufficient, a priority upgrade mechanism is automatically activated to ensure the power supply of core equipment.

[0045] By incrementally updating local parameters of the model based on user-corrected operation data, the entire model can be retrained without retraining, reducing computational overhead and achieving real-time adaptation.

[0046] Example 2 This embodiment targets a "single home office user" scenario. This user is a newly moved-in tenant, with a stay of one week (low data scenario). Their daily electrical equipment includes: computer, router, phone charger, living room lighting, rice cooker, TV, and oven. The scenario is tagged as "working from home on weekdays + leisure on holidays." The core power requirement is to ensure continuous power supply to office equipment (computer, router), with secondary requirements for basic lighting and communication equipment. This embodiment, based on the method of this invention, achieves intelligent scheduling of emergency power supply priorities for this user's residential energy storage.

[0047] like Figure 2As shown, it includes: S1: Small Sample Data Collection and Feature Engineering of User Behavior The small sample data acquisition unit collects the user's device power consumption time-series data, user operation behavior data, and scene label data in a short period of time. The collected small sample data is preprocessed by cleaning and time-series alignment, and then time-series and semantic features are extracted. The time-series features cover peak power consumption periods and start-stop intervals, while the semantic features are obtained through scene label encoding and user operation behavior sequence encoding. Finally, a user behavior small sample feature set is constructed.

[0048] S2: Construction and Training of a User Behavior Few-Shot Learning Model Based on Meta-Learning A model-independent meta-learning algorithm is adopted to construct a two-level framework of "meta-training-fine-tuning". First, an initial meta-model is obtained by training a common electricity consumption behavior dataset of multiple household types. Then, the constructed user behavior small sample feature set is input into the initial meta-model, and the model parameters are adapted through a small number of gradient descent iterations to obtain a personalized user behavior recognition model adapted to the user. The model outputs the user dependence score of each electrical device.

[0049] S3: Initial Priority Ranking of Emergency Power Supply Equipment First, a basic score for the emergency demand of each device is preset based on its functional attributes. Among them, the basic scores of medical, lighting, and communication devices are higher than those of entertainment and non-essential kitchen devices. Then, combined with the user dependence score output by the personalized user behavior recognition model, a two-dimensional priority evaluation function is used to calculate the priority score of each device. Based on the score results, the initial priority ranking of each device is completed, with office and communication devices ranked higher due to their high user dependence.

[0050] S4: Dynamic Priority Adjustment and Dispatch Command Generation in Emergency Situations When the power grid fails and the emergency power supply mode is triggered, the remaining power, output power limit and real-time power demand of each device of the household energy storage system are collected in real time. Based on the initial priority score, a dynamic adjustment model is constructed with the goal of "maximizing the satisfaction of core needs and efficient utilization of energy storage power". The dynamic priority sequence and the corresponding power allocation scheduling instructions are obtained by combining the remaining power status of energy storage to determine whether to activate the priority upgrade mechanism.

[0051] S5: Scheduling Execution and Model Feedback Optimization After receiving the dispatch command, the energy storage dispatch execution unit completes the power supply switching and power allocation to ensure seamless emergency power supply. At the same time, it monitors the user's corrective operation behavior and equipment operating status in real time during the emergency power supply process, and feeds back the collected corrective data to the edge intelligent computing unit. The local parameters of the personalized user behavior recognition model are incrementally updated through the small sample incremental learning algorithm, without the need to retrain the entire model.

[0052] Example 3 like Figure 3 As shown, a user-based energy storage emergency power supply priority intelligent scheduling system integrating user behavior few-sample learning includes: The small sample data acquisition unit is used to collect device power consumption time-series data, user operation behavior data, and scene label data. The edge intelligent computing unit is used to run the user behavior few-sample learning model to realize personalized user behavior recognition and dynamic priority calculation. The energy storage dispatch execution unit is used to receive dispatch instructions and execute power allocation and power supply switching; The feedback interaction unit is used to collect user correction operation data and feed it back to the edge intelligent computing unit. The power switching unit is used to quickly switch to energy storage power supply mode when the power grid fails.

[0053] As a refinement of the above embodiments, the edge intelligent computing unit is equipped with a lightweight meta-learning inference framework, which supports the deployment and inference of model-independent meta-learning algorithms; The energy storage scheduling execution unit includes an energy storage converter control module and a battery management system interaction module.

[0054] The small sample data acquisition unit includes a smart meter, a device power consumption monitoring module, a user operation interaction module, and a scene tag input module, which are used to collect device power consumption time series data, user operation behavior data, and scene tag data.

[0055] The edge intelligent computing unit is equipped with a lightweight meta-learning inference framework, which is used to run a few-sample learning model of user behavior to achieve personalized user behavior recognition and dynamic priority calculation.

[0056] The energy storage scheduling execution unit includes an energy storage converter (PCS) control module and a battery management system (BMS) interaction module, which are used to receive scheduling instructions and execute power allocation and power supply switching.

[0057] Combination Figure 4As shown, this disclosure provides a smart scheduling device 300 for priority emergency power supply of residential energy storage that integrates user behavior few-shot learning, including a processor 304 and a memory 301. Optionally, the device may further include a communication interface 302 and a bus 303. The processor 304, communication interface 302, and memory 301 can communicate with each other via the bus 303. The communication interface 302 can be used for information transmission. The processor 304 can call logical instructions in the memory 301 to execute the smart scheduling method for priority emergency power supply of residential energy storage that integrates user behavior few-shot learning described in the above embodiment.

[0058] Furthermore, the logic instructions in the aforementioned memory 301 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium.

[0059] The memory 301, as a computer-readable storage medium, can be used to store software programs and computer-executable programs, such as the program instructions / modules corresponding to the methods in the embodiments of this disclosure. The processor 304 executes functional applications and data processing by running the program instructions / modules stored in the memory 301, thereby realizing the intelligent scheduling method for priority emergency power supply of residential energy storage that integrates small-sample learning of user behavior in the above embodiments.

[0060] The memory 301 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the terminal device. Furthermore, the memory 301 may include high-speed random access memory and may also include non-volatile memory.

[0061] This disclosure provides a computer-readable storage medium storing computer-executable instructions, which are configured to execute the above-described intelligent scheduling method for priority emergency power supply of residential energy storage that integrates user behavior small-sample learning.

[0062] The aforementioned computer-readable storage medium may be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.

[0063] The technical solutions of this disclosure can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes one or more instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in this disclosure. The aforementioned storage medium can be a non-transitory storage medium, including: a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and other media capable of storing program code; it can also be a transient storage medium.

[0064] The foregoing description and accompanying drawings fully illustrate embodiments of this disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, procedural, and other changes. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the order of operation may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. Moreover, the terminology used in this application is for describing embodiments only and is not intended to limit the claims. As used in the description of embodiments and claims, the singular forms “a,” “an,” and “the” are intended to equally include the plural forms unless the context clearly indicates otherwise. Similarly, the term “and / or” as used in this application refers to any and all possible combinations of one or more of the associated listed elements. Additionally, when used in this application, the term "comprise" and its variations "comprises" and / or "comprising" refer to the presence of stated features, integrals, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof. Without further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the process, method, or apparatus that includes said element. In this document, each embodiment may focus on the differences from other embodiments, and similar or identical parts between embodiments can be referred to mutually. For methods, products, etc., disclosed in the embodiments, if they correspond to the method section disclosed in the embodiments, the relevant parts can be referred to the description of the method section.

[0065] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of this disclosure. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0066] The methods and products (including but not limited to devices and equipment) disclosed in the embodiments herein can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units may be merely a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the coupling or direct coupling or communication connection shown or discussed between each other may be through some interfaces, and the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units may be selected to implement this embodiment according to actual needs. In addition, the functional units in the embodiments of this disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0067] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

Claims

1. A method for prioritizing and intelligently scheduling emergency power supply for residential energy storage based on small-sample learning of user behavior, characterized in that, Includes the following steps: Collect small sample data of users, including device power consumption time series data, user operation behavior data and scene label data, perform data cleaning, time series alignment and feature extraction, and construct a small sample feature set of user behavior; A user behavior few-sample learning model based on meta-learning is constructed. An initial meta-model is trained through a meta-training phase, and the initial meta-model is fine-tuned using the few-sample feature set in a fine-tuning phase to obtain a personalized user behavior recognition model. Based on the personalized user behavior recognition model, the user dependence score of each electrical device is output. Combined with the preset emergency demand score of the device, the initial priority score of each device is calculated through a two-dimensional priority evaluation function and then sorted. In emergency power supply mode, the remaining energy storage capacity, output power limit and power demand of each device are obtained in real time. Based on the initial priority score, a dynamic adjustment model is constructed to generate dynamic priority sequence and corresponding power allocation scheduling instructions. The scheduling instructions are executed, and user correction operation data is collected during the power supply process. The personalized user behavior recognition model is updated through small-sample incremental learning.

2. The intelligent scheduling method for priority emergency power supply of residential energy storage based on user behavior small-sample learning as described in claim 1, characterized in that, The feature extraction includes temporal feature extraction and semantic feature extraction. The temporal features include peak power consumption periods, start-stop intervals, and power fluctuation coefficients. The semantic features are obtained through one-hot encoding of scene tags and sequence encoding of user operation behaviors.

3. The intelligent scheduling method for priority emergency power supply of residential energy storage based on user behavior small-sample learning as described in claim 1, characterized in that, The meta-learning adopts a model-independent meta-learning algorithm, and the meta-training stage uses a multi-household common electricity consumption behavior dataset for training.

4. The intelligent scheduling method for priority emergency power supply of residential energy storage based on user behavior small-sample learning as described in claim 1, characterized in that, The two-dimensional priority evaluation function is: in, Rate the priority. Rate user dependency. Score the equipment's emergency demand level. This is the weighting coefficient, with a value range of 0.4–0.

6.

5. The intelligent scheduling method for priority emergency power supply of residential energy storage based on user behavior small-sample learning as described in claim 1, characterized in that, The constraints of the dynamic adjustment model include: The total output power of energy storage shall not exceed the rated power; The power supply of each device shall not be less than its minimum operating power; The remaining energy storage capacity should not be lower than the emergency protection threshold; When the remaining energy storage capacity is lower than a preset threshold, a priority upgrade mechanism is activated, supplying power only to high-priority devices that are within the previous preset proportion.

6. The intelligent scheduling method for priority emergency power supply of residential energy storage based on user behavior small-sample learning as described in claim 1, characterized in that, The few-sample incremental learning only updates the local parameters of the personalized user behavior recognition model, without retraining the entire model.

7. A priority intelligent scheduling system for residential energy storage emergency power supply that integrates user behavior few-sample learning to implement the method described in any one of claims 1-6, characterized in that, include: The small sample data acquisition unit is used to collect device power consumption time-series data, user operation behavior data, and scene label data. The edge intelligent computing unit is used to run the user behavior few-sample learning model to realize personalized user behavior recognition and dynamic priority calculation. The energy storage dispatch execution unit is used to receive dispatch instructions and execute power allocation and power supply switching; The feedback interaction unit is used to collect user correction operation data and feed it back to the edge intelligent computing unit. The power switching unit is used to quickly switch to energy storage power supply mode when the power grid fails.

8. The intelligent scheduling system for priority emergency power supply of residential energy storage based on user behavior small-sample learning as described in claim 7, characterized in that, The edge intelligent computing unit is equipped with a lightweight meta-learning inference framework, which supports the deployment and inference of model-independent meta-learning algorithms; The energy storage scheduling execution unit includes an energy storage converter control module and a battery management system interaction module.

9. A user-use energy storage emergency power supply priority intelligent scheduling device integrating user behavior few-sample learning, comprising a processor and a memory storing program instructions, characterized in that, The processor is configured to execute, when running the program instructions, the intelligent scheduling method for priority emergency power supply of residential energy storage as described in any one of claims 1-6, which integrates user behavior few-sample learning.

10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the intelligent scheduling method for priority emergency power supply of residential energy storage, which integrates user behavior small-sample learning as described in any one of claims 1-6.

Images